Hierarchical Relationships: A New Perspective to Enhance Scene Graph Generation
It addresses the issue of low annotation quality and limited predicate sets in scene graph generation for computer vision applications, though it presents preliminary findings.
This paper tackles the problem of scene graph generation by leveraging hierarchical structures among labels for relationships and objects, resulting in improved performance on the Visual Genome dataset, particularly in predicate classifications and zero-shot settings.
This paper presents a finding that leveraging the hierarchical structures among labels for relationships and objects can substantially improve the performance of scene graph generation systems. The focus of this work is to create an informative hierarchical structure that can divide object and relationship categories into disjoint super-categories in a systematic way. Specifically, we introduce a Bayesian prediction head to jointly predict the super-category of relationships between a pair of object instances, as well as the detailed relationship within that super-category simultaneously, facilitating more informative predictions. The resulting model exhibits the capability to produce a more extensive set of predicates beyond the dataset annotations, and to tackle the prevalent issue of low annotation quality. While our paper presents preliminary findings, experiments on the Visual Genome dataset show its strong performance, particularly in predicate classifications and zero-shot settings, that demonstrates the promise of our approach.